What is the Supercomputing Internet
The Supercomputing Internet is a national computing power service platform launched under the guidance of the Ministry of Science and Technology. It integrates computing power resources such as supercomputing and intelligent computing across the country to achieve coordinated planning and scheduling of computing power. Based on providing one-stop services including computing power, applications, data, ecosystem, and community, the Supercomputing Internet promotes the inclusiveness and ease of use of computing power. The platform supports online inference, customized development, and API deployment for a variety of AI models (such asDeepSeek, Qwen, etc.), and provides visual interactive services such as Chatbot, helping to drive scientific and technological innovation and high-quality development of the digital economy.
Main functions of the Supercomputing Internet
- Integration and scheduling of computing power resources: Integrates computing power resources such as supercomputing and intelligent computing across the country to achieve coordinated scheduling of computing power.
- Model services: Provides ready-to-use AI model services, supporting online inference and customized development for a variety of models (such as DeepSeek, Qwen, etc.).
- Application software support: Provides coverage of multiple fields including life sciences, industrial design, and quantum computing, meeting the computing needs of all industries.
- SaaS platform services: Adopts an installation-free and maintenance-free computing approach. Users obtain the required resources based on subscription services, lowering the threshold for use.
- Data support: Provides abundant scientific research data and data analysis tools, helping researchers explore new research directions and promoting research innovation.
- Technical services and support: Provides one-stop technical solutions, including model deployment, API calls, Chatbot interaction, etc., and is also equipped with a professional technical team to provide 7×24 support.
Official website address of the Supercomputing Internet
- Official website address：scnet.cn
Deployment and services of the DeepSeek series models
- DeepSeek-R1 full suite launched: For junior developers, providesDeepSeek-R1(1.5B, 7B, 8B, 14B) online inference services. Users can experience it with one click and freely download model files for development.
- R1 model API deployment: For enterprise-level users and senior developers, provides the DeepSeek-R1 model image of up to the 671B full version, supports rapid deployment of API services and AI Web applications, and users can deploy AI applications based on the API call address provided by the platform.
- Chatbot visual dialogue interface: For ordinary users, provides a Chatbot visual interface for the DeepSeek-R1 32B version. Users can converse with the model in real time, and more versions will be updated in the future.
How to perform private deployment of DeepSeek
- Create an environment and project space：
- Create a Notebook online and launch the basic image of heterogeneous accelerator cards, log in to the Supercomputing Internet (scnet.cn) official website, complete the login, and click "Console" in the upper right corner;
- Click "Notebook" in the shortcut entry to enter the Create Notebook page;
- Use the Notebook lightweight development tool for development and debugging：
- In the Notebook of the Supercomputing Internet platform, users select the [Heterogeneous Accelerator AI] resource and choose the corresponding development environment.
- In [Model Image], select the target DeepSeek model image (such as versions 671B, 70B, 32B, 14B, 8B, 7B, 1.5B, etc.).
- After starting the instance, debug according to the Markdown instructions, deploy parameters, or carry out personalized development, supporting debugging based on WebUI or API interface services.
- After debugging is completed, users can save the image for subsequent deployment.
- Rapid Model Deployment：
- Users use the platform's image management function to push images saved in Notebook to the [Model Deployment] service.
- Fill in the API call address and select the [Heterogeneous Accelerator Card AI] resource.
- The platform will quickly deploy the model as an online service and provide an API interface, based on which users can call the model service.
Application Scenarios of the Supercomputing Internet
- Scientific Research and Academia: Supports life sciences, quantum computing, and basic scientific research, providing large-scale simulation and data analysis capabilities.
- Industry and Manufacturing: Assists high-end manufacturing in complex design, simulation analysis, and digital twin system development.
- Enterprise Digitalization: Provides AI model development, big data analysis, and SaaS services to accelerate enterprise intelligent transformation.
- Artificial Intelligence: Supports large model training, inference, and AI application development and deployment.
- Education and Training: Assists university research and teaching and the development of online education platforms.